Non-Parametric Methods
Non-Parametric Methods
Non-parametric methods are procedures of
statistical inference for models without
precise assumptions about the distribution
of the sample.
These methods does not rely on any data
referring to any particular parametric group
of probability distributions.
Mann-Whitney U Test
Mann Whitney U Test is a non-parametric statistical test
used to compare two samples or groups.
Alternative version of the two-sample t-test.
Itis used to examine whether the two groups have the
identical distribution or not.
Mann Whitney U Test is based on ranks of the
independent samples.
Assumptions
1. The dependent variable is numerical variable
2. The independent variable is a nominal
variable. It contains two categories of variable
(grouping variable)
3. The observations are independent
4. The two variables are not normally distributed.
Example
Testing Procedure of the test
Step 1: Setting the Hypothesis
The null and the alternative hypothesis is:
OR
Step 2: Level of significance ()
The level of significance is 1%, 5% or 10%.
Step 3: Test Statistic
The Mann-Whitney test statistic is U-statistic which is defined as:
Where
=Number of observations in first sample
=Number of observations in second sample
and is the sum of rank of sample 1 and sample 2 respectively.
𝑛1 (𝑛1 +1)
𝑈 1=𝑛 1 𝑛2 + − 𝑅1
2
Determination of and :
To carry out the test the two samples are combined in increasing
magnitude and assign ranks from 1,2, …, .
In case of ties the average of tied rank is given
Then, is determined as the sum of rank of observations from
group 1 (sample 1)
Then is determined as the sum of rank of observations from
group 2 (sample 2)
Step 4: Decision Rule
The null hypothesis of the Mann Whitney U test is
rejected in the p-value is less than level of
significance.
Step 5: Computation of Test Statistic
Using sample information test statistic is
computed.
Step 6: Conclusion
Based on Step 4 Decision Rule and Step 5
calculation the conclusion is developed.
Example
Example
Step 1: The null and the alternative hypothesis is:
Step2: Level of significance ()
The level of significance is 5%.
Step 3: Test Statistic
The Mann-Whitney test statistic is U-statistic is:
Where
=Number of observations in first sample
=Number of observations in second sample
and is the sum of rank of sample 1 and sample 2 respectively.
Step 4: Decision Rule
The null hypothesis of the Mann Whitney U test is
rejected in the p-value is less than 0.05.
Step 5: Computation of Test Statistic
Using SPSS
Data Entry Requirements
Data of Variable to be tested
Grouping Variable Codes:
(nominal variable)
Plastic 1=1
Plastic 2= 2
SPSS
Path for
the test
Continued..
?? Require to
define
grouping
Check the boxes for Test and click OK
Output of SPSS Information of
Test
Step 5: Test Statistic
The test statistic is
Where
29
Now
Step 6: Conclusion
As
p-value of U-statistic (0.078)>0.05
Decision:
The null hypothesis is accepted at 5% level of
significance (p-value=0.093>0.05)
Interpretation:
There is no difference in the distribution of two
types of plastics
